cnv_game2fasta.pl

#!/usr/bin/perl -w
#-----------------------------------------------------------+
#                                                           |
# cnv_game2fasta.pl - Convert game.xml to fasta file        |
#                                                           |
#-----------------------------------------------------------+
#                                                           |
#  AUTHOR: James C. Estill                                  |
# CONTACT: JamesEstill_@_gmail.com                          |
# STARTED: 12/18/2007                                       |
# UPDATED: 01/04/2010                                       |
#                                                           |
# DESCRIPTION:                                              |
#  Convert sequence data from game.xml file to the fasta    |
#  file format. Requires XML::Simple                        |
#                                                           |
# USAGE:                                                    |
#  cnv_game2fata.pl -i infle -o outfile                     |
#                                                           |
# VERSION: $Rev$                                            |
#                                                           |
#                                                           |
# LICENSE:                                                  |
#  GNU General Public License, Version 3                    |
#  http://www.gnu.org/licenses/gpl.html                     |  
#                                                           |
#-----------------------------------------------------------+

package DAWGPAWS;

#-----------------------------+
# INCLUDES                    |
#-----------------------------+
use strict;
use Text::Wrap;                # To print the output wrapped 80 characters
use XML::Simple;               # To parse the XML input
use Getopt::Long;              # Get options from the command line
# The following needed for printing help
use Pod::Select;               # Print subsections of POD documentation
use Pod::Text;                 # Print POD doc as formatted text file
use IO::Scalar;                # For print_help subfunction
use IO::Pipe;                  # Pipe for STDIN, STDOUT for POD docs
use File::Spec;                # Convert a relative path to an abosolute path

#-----------------------------+
# PROGRAM VARIABLES           |
#-----------------------------+
my ($VERSION) = q$Rev$ =~ /(\d+)/;

#-----------------------------+
# VARIABLE SCOPE              |
#-----------------------------+
my $infile;
my $outfile;

# BOOLEANS
my $quiet = 0;
my $verbose = 0;
my $show_help = 0;
my $show_usage = 0;
my $show_man = 0;
my $show_version = 0;
my $do_test = 0;                  # Run the program in test mode

# Set the text wrap columns to 60
$Text::Wrap::columns = 60;

#-----------------------------+
# COMMAND LINE OPTIONS        |
#-----------------------------+
my $ok = GetOptions(# REQUIRED OPTIONS
		    "i|infile=s"  => \$infile,
                    "o|outfile=s" => \$outfile,
		    # ADDITIONAL OPTIONS
		    "q|quiet"     => \$quiet,
		    "verbose"     => \$verbose,
		    # ADDITIONAL INFORMATION
		    "usage"       => \$show_usage,
		    "test"        => \$do_test,
		    "version"     => \$show_version,
		    "man"         => \$show_man,
		    "h|help"      => \$show_help,);

#-----------------------------+
# SHOW REQUESTED HELP         |
#-----------------------------+


if ( ($show_usage) ) {
#    print_help ("usage", File::Spec->rel2abs($0) );
    print_help ("usage", $0 );
}

if ( ($show_help) || (!$ok) ) {
#    print_help ("help",  File::Spec->rel2abs($0) );
    print_help ("help",  $0 );
}

if ($show_man) {
    # User perldoc to generate the man documentation.
    system ("perldoc $0");
    exit($ok ? 0 : 2);
}

if ($show_version) {
    print "\ncnv_game2fasta.pl:\n".
	"Version: $VERSION\n\n";
    exit;
}

#-----------------------------------------------------------+
# MAIN PROGRAM BODY                                         |
#-----------------------------------------------------------+


#-----------------------------+
# FILE IO                     |
#-----------------------------+
# If not outfile given, append fasta to infile name
unless ($outfile) {
    $outfile = $infile.".fasta";

    # Will use a hext short name for now
    # this only works with the full path is not being used
    my $infile_prefix = substr ( $infile , 0, 10 );
    $outfile = $infile_prefix.".fasta";
}


open (FASTOUT, ">$outfile") ||
    die "Could not open the output file:\n$outfile\n";

# create the xml object
my $game_xml = new XML::Simple;

# read the xml file
my $game_data = $game_xml->XMLin($infile);


#-----------------------------+
# DATA DUMP FOR TESTING       |
#-----------------------------+
#use Data::Dumper;
#print STDERR Dumper($game_xml);
#exit;


# lets take a look at what we have here
#use Data::Dumper;
#print STDOUT Dumper($game_data);

#print STDERR "version: ".$game_data->{version}."\n";
my $seq_info = $game_data->{seq};
my $seq_string = $seq_info->{residues};
my $seq_id = $seq_info->{name};
#print STDERR "Seq string".$seq_string."\n";
#print STDERR "seqe".$game_data->{seq->{residues}}."\n";
#print STDERR $game_data->{residues};

# remove blank lines
$seq_string =~ s/\n+/\n/g;
# remove white spaces
$seq_string =~ s/\s+//g; 

#-----------------------------+
# PRINT THE FASTA OUTPUT FILE |
#-----------------------------+
# will want to print 80 characters wide
print STDERR $seq_id."\n";
print FASTOUT ">".$seq_id."\n";
#print FASTOUT $seq_string."\n";
print FASTOUT wrap('','', $seq_string);

close (FASTOUT);

exit 0;


#-----------------------------------------------------------+ 
# SUBFUNCTIONS                                              |
#-----------------------------------------------------------+


sub print_help {
    my ($help_msg, $podfile) =  @_;
    # help_msg is the type of help msg to use (ie. help vs. usage)
    
    print "\n";
    
    #-----------------------------+
    # PIPE WITHIN PERL            |
    #-----------------------------+
    # This code made possible by:
    # http://www.perlmonks.org/index.pl?node_id=76409
    # Tie info developed on:
    # http://www.perlmonks.org/index.pl?node=perltie 
    #
    #my $podfile = $0;
    my $scalar = '';
    tie *STDOUT, 'IO::Scalar', \$scalar;
    
    if ($help_msg =~ "usage") {
	podselect({-sections => ["SYNOPSIS|MORE"]}, $0);
    }
    else {
	podselect({-sections => ["SYNOPSIS|ARGUMENTS|OPTIONS|MORE"]}, $0);
    }

    untie *STDOUT;
    # now $scalar contains the pod from $podfile you can see this below
    #print $scalar;

    my $pipe = IO::Pipe->new()
	or die "failed to create pipe: $!";
    
    my ($pid,$fd);

    if ( $pid = fork() ) { #parent
	open(TMPSTDIN, "<&STDIN")
	    or die "failed to dup stdin to tmp: $!";
	$pipe->reader();
	$fd = $pipe->fileno;
	open(STDIN, "<&=$fd")
	    or die "failed to dup \$fd to STDIN: $!";
	my $pod_txt = Pod::Text->new (sentence => 0, width => 78);
	$pod_txt->parse_from_filehandle;
	# END AT WORK HERE
	open(STDIN, "<&TMPSTDIN")
	    or die "failed to restore dup'ed stdin: $!";
    }
    else { #child
	$pipe->writer();
	$pipe->print($scalar);
	$pipe->close();	
	exit 0;
    }
    
    $pipe->close();
    close TMPSTDIN;

    print "\n";

    exit 0;
   
}

1;
__END__

=head1 NAME

cnv_game2fasta.pl - Convert game.xml to fasta format file.

=head1 VERSION

This documentation refers to program version $Rev$

=head1 SYNOPSIS

=head2 Usage

    cnv_game2fasta.pl -i infile.game.xml -o outfile.fasta

=head2 Required Arguments

    --infile        # Path to the input file
    --outfie        # Path to the output file

=head1 DESCRIPTION

This is what the program does

=head1 REQUIRED ARGUMENTS

=over 2

=item -i,--infile

Path of the input file.

=item -o,--outfile

Path of the output file.

=back

=head1 OPTIONS

=over 2

=item --usage

Short overview of how to use program from command line.

=item --help

Show program usage with summary of options.

=item --version

Show program version.

=item --man

Show the full program manual. This uses the perldoc command to print the 
POD documentation for the program.

=item -q,--quiet

Run the program with minimal output.

=back

=head1 EXAMPLES

The following are examples of how to use this script

=head2 Typical Use

This is a typcial use case.

=head1 DIAGNOSTICS

=over 2

=item * Expecting input from STDIN

If you see this message, it may indicate that you did not properly specify
the input sequence with -i or --infile flag. 

=back

=head1 CONFIGURATION AND ENVIRONMENT

Names and locations of config files
environmental variables
or properties that can be set.

=head1 DEPENDENCIES

Other modules or software that the program is dependent on.

=head1 BUGS AND LIMITATIONS

Any known bugs and limitations will be listed here.

=head1 REFERENCE

A manuscript is being submitted describing the DAWGPAWS program. 
Until this manuscript is published, please refer to the DAWGPAWS 
SourceForge website when describing your use of this program:

JC Estill and JL Bennetzen. 2009. 
The DAWGPAWS Pipeline for the Annotation of Genes and Transposable 
Elements in Plant Genomes.
http://dawgpaws.sourceforge.net/

=head1 LICENSE

GNU General Public License, Version 3

L<http://www.gnu.org/licenses/gpl.html>

=head1 AUTHOR

James C. Estill E<lt>JamesEstill at gmail.comE<gt>

=head1 HISTORY

STARTED:

UPDATED:

VERSION: $Rev$

=cut

#-----------------------------------------------------------+
# HISTORY                                                   |
#-----------------------------------------------------------+
#
# 01/04/2010
# - Using xml::simple, only about 30% of the game.xml files
#   are being properly parsed. Working on alternative methods.